Abstract: Product introduction plays an important role in consumer purchase decisions. It is of practical significance to understand how to design and craft the text introduction so as to study consumer purchase decision making behavior of such products. However, previous marketing text analysis researches have mainly focused on consumer-side variables (e.g., online reviews), and didn't pay enough attention on producer-side variables (e.g., product introduction). Also, the outcome variables in previous text analysis studies included only non-purchase variables such as shares, comments and likes. Due to the difficulty of matching product sales at an aggregate level, few studies have used actual secondary data to quantify the impact of text-based outcomes.To overcome the mentioned research gap, based on relevant marketing practices and heuristic-systemic model, two characteristics of movie introduction (readability and attractiveness) are proposed. Then, used a machine learning SVM model, the readability and attractiveness level of the introduction of 264 films released in mainland China in 2018 are computed and its impact on the box office in short term and long term are tested. It is found that: movie introduction’s readability and attractiveness have a different impact on first week’s box office and total box office. Readability has a greater impact on first week’s box office, and attractiveness has a greater impact on total box office. Beside, the presence of stars undermined the impact of readability and attractiveness on the box office, and readability and attractiveness have significant negative interactions effect on first week’s box office and total box office in the high-rating movie group. The research conclusion has great practical significance for the development of the Chinese movie industry and reasonable guidance for movie managers in promoting movie sales.

Key words: movie introduction; readability; attractiveness; machine learning